Companies Must Disclose Basis for AI-Driven Layoffs
Most provisions of Connecticut's AI regulatory law, the 'Connecticut AI Accountability and Transparency Act (CART Act),' took effect on October 1 (local time). The 39-article bill, the product of compromise between the state Senate, House, and governor, was signed by Governor Ned Lamont on May 27. Companies conducting mass layoffs must report to the state labor department whether the layoffs are related to AI use or technological change. The excuse that 'the computer made them do it' no longer flies.
The key change is in the employment sector. Employers issuing WARN Act notices for mass layoffs or plant closures must submit information on the causes of the layoffs to the Connecticut Department of Labor. Companies can also no longer use automation as a defense when hiring and employment technologies result in discrimination. The law explicitly states that the use of automated employment decision technology cannot serve as a defense against discrimination claims.
Consumer protection provisions also took effect. Generative AI providers with more than 1 million monthly users that sell subscription products in Connecticut must clearly state key terms such as usage limits and blocking capabilities, and customers must agree in writing when signing up or renewing. Major providers must embed information that identifies images, audio, and video as computer-generated.
State government agencies now face standards as well. Agencies using AI in functions that could significantly affect residents' rights, safety, and welfare must follow state standards, and agencies with the authority to procure AI must complete impact assessments. Whistleblower protections also took effect the same day. Retaliation against whistleblowers is prohibited, with civil penalties of up to $1,000 per violation.
From January 1, 2027, safety obligations will apply to operators of AI companions (chatbots). Operators must implement safeguards that prevent responses encouraging self-harm or violence, maintain procedures to detect signs of suicide risk, self-harm, or imminent violence, and connect users to suicide prevention counseling resources. Chatbots must disclose that they are not human. For minor users, encouraging self-harm, discouraging help-seeking, romantic or sexual interactions, and manipulative engagement tactics are prohibited.
The core employment obligations begin on October 1, 2027. Employers using AI tools in hiring must provide written notice to applicants stating that the technology is being used, its purpose, and the scope and sources of personal data analyzed. On January 1, 2028, youth social media regulations will take effect, restricting the use of minors' activity records to determine feed content without parental consent. Enforcement authority rests mostly with the state Attorney General's office.
